A computer that can function like a human brain is not just the stuff of science fiction, but also the dream of many scientists and engineers. Now, IBM has taken the first step towards making that dream a reality with its TrueNorth superprocessor, which mimics the human brain's abilities and power efficiency.
Standard processors use series of mathematical calculations to solve problems. IBM created TrueNorth, however, to solve problems by understanding its environment, including deriving meaning through context, even with ambiguous topics.
IBM researchers describe the chip as working like the human brain by recognizing patterns and sending neurons across a neural network to compute data. Most impressively, even with over 5 billion transistors, 1 million neurons and 256 synapses, TrueNorth is capable of doing this with just 70 milowatts of power. In comparison, most modern processors have a little over one billion transistors but need up to 140 watts of power.
TrueNorth is part of IBM's larger SyNapse cognitive computing program. IBM suggests that the chip could easily be integrated for technology that requires this sort of real world thinking, such as self-driving cars and smarter smartphones.
"It is a remarkable achievement in terms of scalability and low power consumption," says Horst Simon, deputy director of the Lawrence Berkeley National Laboratory. "The IBM SyNapse project is an indicator of that change that will happen in the next 10 years."
IBM is no stranger to making smarter computers. The company's Watson supercomputer beat the best "Jeopardy" players in the world in 2011. Even Watson, though, has its limits. Not only does it still get stumped by ambiguous questions, but it also requires a massive amount of power.
However, because of TrueNorth's design, it can see through more ambiguous data.
IBM's announcement has faced skepticism, especially from Facebook's director of artificial intelligence, Yann LeCunn. He claims that the chip's function is extremely limited and that its technology is unimpressive.
Although this might be true now, others say that the technology behind TrueNorth just needs time.
"The TrueNorth chip is like the first transistor," says Terrence J. Sejnowski, Salk Instute's Computational Neurobiology Laboratory director. "It will take many generations before it can compete, but when it does, it will be a scalable architecture that can be delivered to cellphones, something that Yann's G.P.U.s will never be able to do."
IBM developed TrueNorth in a partnership with the U.S. Defense Advanced Research Agency (DARPA), so we can expect to see this technology's implementation in the country's defense systems first, perhaps providing better detection of incoming threats. However, the technology isn't strictly limited to military use: it could make for smarter robots and self-driving cars, too.
